In this thread over in "Chip Amps":
http://www.diyaudio.com/forums/showthread.php?postid=1445191#post1445191
Tom Gootee wrote:
I thought I had all this down, but this but I must have missed something along the way. NP0 caps have better tempco, and better tolerance than X7R types, right? So why not use them for supply decoupling?
Tom...you said it, can you educate me?
--Greg
http://www.diyaudio.com/forums/showthread.php?postid=1445191#post1445191
Tom Gootee wrote:
I thought I had all this down, but this but I must have missed something along the way. NP0 caps have better tempco, and better tolerance than X7R types, right? So why not use them for supply decoupling?
Tom...you said it, can you educate me?
--Greg
Can't speak for Tom, but IMO NP0 caps are nearly perfect. Almost no dielectric absorption, almost no dissipation factor, stable, etc. The only problem is they only come in small values. If you need something larger, a different type cap is more practical. One caveat- the very low DF translates into very high Q (1/DF). If you have a situation where lead inductance is significant, you can get ringing. For bypass caps right at the IC pins, this shouldn't be an issue. The bypass cap test written by Jim Williams of Linear Technology and published in their app notes is always a good read when thinking about this.
I think it's as usual a rumour. In many place (not all), ceramics are excellent. Recently I have worked with TPS61031 boost converter and tantalum was not very exciting but 4 x 47 uF X5R did create an excellent output voltage at 600 kHz switch frequency. Still, the tanalum caps was Vishay low ESR but the X5R was much better. Ceramics has many not so good properties but if you know the material, you know also where to use them.
I think what is being said here is that in some locations a perfect capacitor can be a bad thing. Check some of Eva's post for ringing evidence.
Adopting a less than perfect capacitor can give better performance if the correct imperfections are selected.
Gootee is suggesting that x7r has worked for him and could be worth trying.
Adopting a less than perfect capacitor can give better performance if the correct imperfections are selected.
Gootee is suggesting that x7r has worked for him and could be worth trying.
Hi Greg,
I just finally saw this thread. And I see that AndrewT 'beat me to the punch', yet again.
I don't consider myself to be an expert. But I will try to explain a little bit more about what I was thinking, when I posted that.
Yes, just as Andrew has said, my suggestion was aimed at the mere possibilty that a cap that was 'too good' might cause a problem, possibly exciting unwanted high-frequency resonance modes, when paralleled with an electrolytic. Therefore, in the absence of sophisticated measurements and calculations, and assuming the goal is for it to work the first time and further experimentation is not desired, a 'lossier' cap might be a better choice (i.e. a better guess). At the time, I probably mainly had in mind the following thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=106648 . I probably wouldn't blindly suggest a film cap, there, either.
I do certainly agree that C0G (aka NPO) caps are really-great caps. I have liked them even more, ever since I read Bob Pease's article about dielectric absorption (aka 'soakage'). I use them a lot, but almost never for power supply decoupling, unless they're part of a snubber network (i.e. resistance added).
I just finally saw this thread. And I see that AndrewT 'beat me to the punch', yet again.
I don't consider myself to be an expert. But I will try to explain a little bit more about what I was thinking, when I posted that.
Yes, just as Andrew has said, my suggestion was aimed at the mere possibilty that a cap that was 'too good' might cause a problem, possibly exciting unwanted high-frequency resonance modes, when paralleled with an electrolytic. Therefore, in the absence of sophisticated measurements and calculations, and assuming the goal is for it to work the first time and further experimentation is not desired, a 'lossier' cap might be a better choice (i.e. a better guess). At the time, I probably mainly had in mind the following thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=106648 . I probably wouldn't blindly suggest a film cap, there, either.
I do certainly agree that C0G (aka NPO) caps are really-great caps. I have liked them even more, ever since I read Bob Pease's article about dielectric absorption (aka 'soakage'). I use them a lot, but almost never for power supply decoupling, unless they're part of a snubber network (i.e. resistance added).
Wow...that's a lot of information to digest, especially in that other thread. I don't have access to a 'scope, and had been using film caps for supply decoupling, without any obvious bad effects. Perhaps I will switch X7R's for that application, since that seems like it might bea genearlly-accepted better choice.
--Greg
--Greg
Linear Technology, AN-13, page 25. Required reading and testing for anybody blindly paralleling caps.
about halfway down...
about halfway down...
NP0/COG caps are common up to around 1nF. For higher
capacitance basically everything is X7R or even X5R for higher capacitance is small size (for example 1uF in 0402 or 10uF in 0603).
At audio frequencies there is basically no difference between a COG and an X7R dielectric. COG have better temperature stability but if these caps are used for decoupling you don't need to have an exact capacitance value.
100nF caps in X7R are a common practice for the decoupling of op-amps and others low power chips.
You start to see differences between COG and X7R only at RF frequencies, let's say above 100MHz. This is due to the higher stray inductance of X7R and the higher dielectric losses at high frequency.
capacitance basically everything is X7R or even X5R for higher capacitance is small size (for example 1uF in 0402 or 10uF in 0603).
At audio frequencies there is basically no difference between a COG and an X7R dielectric. COG have better temperature stability but if these caps are used for decoupling you don't need to have an exact capacitance value.
100nF caps in X7R are a common practice for the decoupling of op-amps and others low power chips.
You start to see differences between COG and X7R only at RF frequencies, let's say above 100MHz. This is due to the higher stray inductance of X7R and the higher dielectric losses at high frequency.
AndrewT said:
What is missing from AN-13? And where did you find it? The bypass-caps section is on page 25, in my just-downloaded copy of AN-13.
-----
There is some more good stuff about bypass caps in AN-47, starting on page 11. Lots of other good stuff in there, too.
And there's some more on page 27 of AN-42, which also has tons of voltage reference and regulator circuits.
For HOW and WHERE to connect the caps, there's the classic AN-202 at analog.com . AN-345, there, is also a must-read.
Not about bypassing, but AN-21 at linear.com is very interesting. It's about composite amplifiers.
All this points out the inherent danger in the audiophile "fetish" of blindly adding "better" bypass caps to commercial gear. But note that most of the so-called problems with "too good" caps is not in the audio band, but out at many MHz or many hundreds of MHz. Looking at various capacitor combinations on an HP/Agilent component analyzer is educational
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.